Much worse performance/compatibility on Ubuntu 17.10 vs Windows 10 w/8700k
#21
Xeno.. pressing f4 toggles between normal and max, pressing tab toggles between turbo and normal, if you are in turbo and hit f4, it toggles between turbo and max
amd tr 1950x                                                 amd ryzen 5 2500u
asrock x399 taichi                                         amd vega 8 
XFX Radeon rx570                                       16gb ddr 4 2400 ram
32gb gskill ddr4-3200                                   1tb nvme ssd
Debian Bullseye                                           480gb sata ssd
Custom loop water cooled                           HP envy x360
nzxt 340 case
Reply

Sponsored links

#22
(02-09-2018, 03:29 PM)gregory Wrote: Ok, Driver is fine. I have no clue.

Do you think buying a dedicated graphics card like a 1050 ti would have any affect on performance in software mode?
Reply
#23
(02-10-2018, 12:48 PM)xenphor Wrote: Do you think buying a dedicated graphics card like a 1050 ti would have any affect on performance in software mode?

Software mode is CPU dependent - the speed it runs relies on your CPU's STR (single thread rating); it doesn't take advantage of a dedicated gpu like hardware mode will.
Reply
#24
Is there anyone else with an 8700k or similar coffeelake processor that is having speed issues with pcsx2 in Linux?
Reply
#25
Define "speed issues" - that can mean alot of things Wink
Reply
#26
(02-12-2018, 03:15 AM)CK1 Wrote: Define "speed issues" - that can mean alot of things Wink

It runs at about half the speed or less than the Windows version according to the counter displayed in the pcsx2 window.
Reply
#27
open source vs. close source drivers can do a lot for video..
What are your freq governors set to?
What is your actual cpu usage and ram usage when playing?
Half is a large decrease, especially if you're using the onboard 630 video card (its the 630 right?) My 620 is faster in windows, but not 100% faster I see a 10-15% decrease based on the % fps counter in the window.

For the more linux savvy folks.. could this be Page Table accessibility changes causing a slowdown on a barely acceptable video card (to mitigate spectre was it?)

Are you running the most recent mesa drivers (I believe I saw yes.. you were.. could be wrong) and processor microcode in your distro?
amd tr 1950x                                                 amd ryzen 5 2500u
asrock x399 taichi                                         amd vega 8 
XFX Radeon rx570                                       16gb ddr 4 2400 ram
32gb gskill ddr4-3200                                   1tb nvme ssd
Debian Bullseye                                           480gb sata ssd
Custom loop water cooled                           HP envy x360
nzxt 340 case
Reply
#28
(02-12-2018, 10:59 AM)scythefwd Wrote: open source vs. close source drivers can do a lot for video..
What are your freq governors set to?
What is your actual cpu usage and ram usage when playing?
Half is a large decrease, especially if you're using the onboard 630 video card (its the 630 right?)  My 620 is faster in windows, but not 100% faster I see a 10-15% decrease based on the % fps counter in the window.

For the more linux savvy folks.. could this be Page Table accessibility changes causing a slowdown on a barely acceptable video card (to mitigate spectre was it?)

Are you running the most recent mesa drivers (I believe I saw yes.. you were.. could be wrong) and processor microcode in your distro?

Well I was just told that gpu has no effect in software mode. I've set the governor to performance and there was no difference. I don't see why ram would be an issue since I have 8gb and have nothing else open in the background. Using performance mode forced my cpu to run at max frequency and it made no difference when playing. I'm using Mesa 18 from the Ubuntu PPA.
Reply
#29
It has a limited effect.. its not processing much, thats all handled in the cpu.. but your gpu shares ram, cache, and cycles with the cpu.. The extra load is minimal, but potentially noticeable when using the integrated gpu.

8gb really isnt a lot of ram, my laptop doubles that, my desktop quadruples that.. Background processes can hog more ram than you think.. I'm assuming you are using a vanilla install, so we're not looking at a sql server in the background, or a samba AD domain on it.. but it doesnt take much.. especially if there is a memory leak.

Do you have ffx2? On max limiter, what fps do you see in OpenGL SW mode in the opening FMV? I'd expect to see you hit a pretty steady 70 fps.. my weakling little 7200u, which has a stp of about 1800.. is pushing 50 - 60fps pretty much throughout the whole fmv. Your 8700k STP is around 2700+.. basically a full 900 higher, 150% faster than mine... If you're not pulling a steady 60 fps in standard software mode in just about any game.. you have a system configuration issue..
amd tr 1950x                                                 amd ryzen 5 2500u
asrock x399 taichi                                         amd vega 8 
XFX Radeon rx570                                       16gb ddr 4 2400 ram
32gb gskill ddr4-3200                                   1tb nvme ssd
Debian Bullseye                                           480gb sata ssd
Custom loop water cooled                           HP envy x360
nzxt 340 case
Reply
#30
ok, just went and checked.. I'm using 5gb of ram, roughly (if I'm reading this right) in use.. and the only thing running at that time was terminal and top, and my base system (fedora 27 workstation)

TOP says I have 16288396 KiB Mem (16gb) total, and 11367264 KiB free, 2166944 used, (that looks like 4-5 gb usage just running the system). 3gb of that not free but not in use is cache. I'm not sure if that is addressible or not to the system.

Conky says its only 2.25GiB in use.. I'm not sure which to trust.. but I'd check top when you're playing on a separate screen if possible.. you may see you're swapping..

Nice thing, running pcsx2 doesnt seem to appreciably effect that ram usage.. < .4gb difference really.
With MTVU hack enabled and VU cycle stealing set to 1, I'm seeing ~70% cpu usage overall (4 threads, dual core).. which is right in line with 3 threads running at 98% as reported by pcsx2
amd tr 1950x                                                 amd ryzen 5 2500u
asrock x399 taichi                                         amd vega 8 
XFX Radeon rx570                                       16gb ddr 4 2400 ram
32gb gskill ddr4-3200                                   1tb nvme ssd
Debian Bullseye                                           480gb sata ssd
Custom loop water cooled                           HP envy x360
nzxt 340 case
Reply




Users browsing this thread: 1 Guest(s)